Research On Fatigue Prediction Algorithm Of EMG Signal Data Fusion Processing

With the increasing improvement of material and living standards,national fitness exercise has become a trend.Although fitness exercises can effectively exercise human muscles,long-term exercise will cause body fatigue,and excessive fatigue will have many adverse effects on the human body,and even directly damage the body's health,so it is of great significance for fatigue assessment and prediction.At present,most of the traditional exercise muscle fatigue assessment research focuses on medical,rehabilitation,sports and other professional fields.There are relatively few researches in the field of daily fitness exercise.Therefore,this paper selects two types of exercises: rope skipping and plank support as the research objects,using multi-sensor data fusion Technology,combined with wavelet transform,Kalman filter,least square method and BP neural network and other algorithms to analyze and study the human body fitness movement.The experimental design of surface EMG signal acquisition for rope skipping and plank support.By analyzing the characteristics of the surface EMG signal,the acquisition experiment equipment is selected,and the scientific experiment design is carried out according to the human muscles,the main exercise parts and the people suitable for exercise during the exercise process,and the experimental data is obtained to provide reliable data support for subsequent analysis.The data fusion algorithm and prediction algorithm are researched and analyzed,the data fusion technology and the fatigue prediction algorithm are combined,and the data fusion muscle fatigue prediction system based on multi-sensor is designed for the two selected sports.Discrete wavelet transform is used to decompose and reconstruct the six-channel surface electromyography(s EMG)signal collected by rope skipping to reduce noise,and the s EMG signal preprocessed by wavelet transform is used to perform the root mean square value(RMS)of the time domain index,Integral electromyography(i EMG)and frequency domain index median frequency(MF),average power frequency(MPF)feature extraction,to achieve feature layer data fusion,and comprehensive muscle active frequency interval analysis to obtain muscle fatigue results;finally use Least squares regression analysis performs data fitting on frequency domain indicators,classifies human body fatigue through frequency domain indicators,and analyzes the results of the least squares method to determine human body fatigue.For the six-channel sEMG signal collected by the plank support movement,the Kalman filter method is used for noise reduction processing,and the time domain index RMS,i EMG and frequency domain index MF,MPF feature extraction are performed on the s EMG signal preprocessed by the Kalman filter to achieve Feature layer data fusion,and comprehensive analysis of muscle activity frequency range to obtain muscle fatigue results;use BP neural network to predict frequency domain feature values ??MF and MPF,combine LM algorithm,RPROP algorithm,SCG algorithm and Bayesian algorithm four An optimization algorithm optimizes the designed neural network;finally,the BP neural network combined with the LM optimization algorithm is used to identify the fatigue degree of all frequency domain features MF and MPF values.The research results show that the design of the muscle fatigue prediction system based on multi-sensor data fusion can effectively obtain the fatigue analysis of sports muscles,and can realize the prediction and evaluation of fatigue.The above research has certain theoretical value and practical value,and can provide quantitative data as a guide for daily fitness exercisers.